PET/MR Imaging in the Detection and Characterization of Pulmonary Lesions: Technical and Diagnostic Evaluation in Comparison to PET/CT

Abstract

Fully integrated PET/MR imaging holds great promise as a novel hybrid imaging modality in oncology and might offer advantages to PET/CT in many instances, especially because of the superior soft-tissue contrast of MR imaging, compared with CT. However, lung metastases are a frequent finding in oncologic patients, and for imaging of the lung CT is still the modality of choice. Thus, we prospectively evaluated differences in quality, detection rate, size, and radiotracer uptake of pulmonary lesions in (18)F-FDG PET/CT and PET/MR imaging. Institutional review board approval and informed consent were obtained. Forty patients (23 men, 17 women; mean age ± SD, 53.2 ± 13.1 y) underwent a single-injection dual-imaging protocol with (18)F-FDG PET/CT and PET/MR imaging. Pulse sequences for the lung included T1-weighted VIBE (volumetric interpolated breath-hold examination) Dixon for attenuation correction and contrast-enhanced VIBE pulse sequences. All patients underwent a diagnostic CT of the chest in deep inspiration, which also served as a standard of reference. Two masked readers assessed in consensus all images randomly concerning quality, detection, standardized uptake value (SUV), and size of pulmonary nodules. Correlations were performed using linear correlation. Overall, 47 pulmonary lesions (mean size ± SD, 10.0 ± 11.4 mm; range, 2-60 mm) in 25 of 40 patients were detected. The PET datasets of PET/MR imaging and PET/CT revealed 22 of 47 pulmonary lesions with focal (18)F-FDG uptake. SUVs of lung lesions in PET/MR imaging and PET/CT correlated significantly (R = 0.9; P = 0.0001) and showed no significant difference (mean SUV PET/MR imaging, 6.3; PET/CT, 5.1; P = 0.388). There was a significantly lower image quality comparing Dixon and VIBE sequence with CT whereas PET from PET/CT and PET from PET/MR imaging showed the same results (2.8). Dixon images detected 15 of 47 lung lesions whereas VIBE images detected 32 of 47 lesions, respectively. The detection rates for small lung lesions less than 1 cm in diameter (n = 33) of MR imaging was significantly lower, with a detection rate of 9 of 33 for the Dixon sequence and 15 of 33 for the VIBE sequence (P < 0.0001 for VIBE and Dixon sequence). There was a high correlation of pulmonary lesion size of CT versus VIBE (R = 0.97). PET image quality and detection rate of (18)F-FDG-positive lung lesions in PET/MR imaging is equivalent to PET/CT despite differences in attenuation-correction techniques. Additionally, a high linear correlation coefficient in the SUVs for the PET images from PET/CT and PET/MR imaging was found. The detection rate of lung lesions can be significantly improved by adding a diagnostic contrast-enhanced VIBE sequence to the PET/MR imaging protocol. However, the detection rate of small lung lesions is still inferior, compared with PET/CT with diagnostic CT of the chest.